Reducing brassinosteroid signalling enhances grain yield in semi-dwarf wheat

Modern green revolution varieties of wheat (Triticum aestivum L.) confer semi-dwarf and lodging-resistant plant architecture owing to the Reduced height-B1b (Rht-B1b) and Rht-D1b alleles1. However, both Rht-B1b and Rht-D1b are gain-of-function mutant alleles encoding gibberellin signalling repressors that stably repress plant growth and negatively affect nitrogen-use efficiency and grain filling2–5. Therefore, the green revolution varieties of wheat harbouring Rht-B1b or Rht-D1b usually produce smaller grain and require higher nitrogen fertilizer inputs to maintain their grain yields. Here we describe a strategy to design semi-dwarf wheat varieties without the need for Rht-B1b or Rht-D1b alleles. We discovered that absence of Rht-B1 and ZnF-B (encoding a RING-type E3 ligase) through a natural deletion of a haploblock of about 500 kilobases shaped semi-dwarf plants with more compact plant architecture and substantially improved grain yield (up to 15.2%) in field trials. Further genetic analysis confirmed that the deletion of ZnF-B induced the semi-dwarf trait in the absence of the Rht-B1b and Rht-D1b alleles through attenuating brassinosteroid (BR) perception. ZnF acts as a BR signalling activator to facilitate proteasomal destruction of the BR signalling repressor BRI1 kinase inhibitor 1 (TaBKI1), and loss of ZnF stabilizes TaBKI1 to block BR signalling transduction. Our findings not only identified a pivotal BR signalling modulator but also provided a creative strategy to design high-yield semi-dwarf wheat varieties by manipulating the BR signal pathway to sustain wheat production.

The green revolution in the 1960s has markedly increased cereal crop yield through widespread cultivation of semi-dwarf and lodgingresistant varieties 1,6 . The beneficial semi-dwarf plant architecture of these green revolution varieties (GRVs) is mainly conferred by the introduction of either of the Reduced height-1 (Rht-1) alleles (Rht-B1b or Rht-D1b) that derived from a gain-of-function mutation of Rht-B1a in the B genome or Rht-D1a in the D genome of wheat (Triticum aestivum L., 2n = 6x = 42, AABBDD genome), and a recessive mutant semi-dwarf1 (sd1) in rice (Oryza sativa L., 2n = 2x = 24). The Rht-B1b, Rht-D1b and sd1 alleles lead to high levels of accumulation of DELLA proteins that repress gibberellin (GA) signalling and further attenuate GA-promoted plant growth to shape semi-dwarfism 1,6 . However, these green revolution alleles also reduce nitrogen (N)-use efficiency (NUE) and carbon fixation, resulting in decreased biomass, spike size and grain weight in the GRVs [3][4][5][6] . Therefore, the GRVs require extremely high N fertilizer inputs to maintain their high yields, but high N input is detrimental to both environments and agriculture sustainability 7 . Identifying new genetic sources that produce desirable semi-dwarf plant architecture with improved NUE without plant growth and grain yield penalty is an urgent goal for continuous improvement of yields of cereal crops in the limited arable lands to feed a growing world population.
Previous studies in rice have established essential roles of the N-regulated plant-specific transcription factor GROWTH-REGULATING FACTOR 4 (GRF4) together with its coactivator GRF-INTERACTING FACTOR1 (GIF1) in activating multiple N-metabolism genes. DELLA proteins inhibit the GRF4-GIF1 activity 2,8 ; however, increasing the abundance of GRF4 can repress DELLA activity to boost NUE and increase biomass and final grain yields in rice and wheat GRVs 2,9,10 . A recent study revealed that an N-induced APETALA2-domain-containing NITROGEN-MEDIATED TILLER GROWTH RESPONSE 5 (NGR5) is a key regulator for genome-wide transcriptional reprogramming in response to N fertilization, and increased NGR5 expression in rice enhanced NUE and grain yield 4 . These studies suggest feasibility to design improved GRVs in cereal crops using the available green revolution genes.
BR has diverse roles in regulating important agronomic traits including plant architecture, spike and panicle morphology, and grain size and shape in cereal crops [10][11][12][13] . BR-deficient crops usually exhibit a dwarf and compact plant stature that is beneficial to lodging resistance and high-density planting [13][14][15] . Here we report a strategy to breed new wheat GRVs with more compact semi-dwarf plant architecture, improved NUE and enhanced grain yields using a rare, natural deletion of a haploblock, designated as r-e-z. The deleted haploblock includes

Identification of the rare haploblock deletion
Analysis of quantitative trait loci in a segregating wheat population of Heng597 (Heng) × Shi4185 (Shi) identified a quantitative trait locus, QTgw.cau-4B, for higher thousand-grain weight (TGW) from Heng (Fig. 1a, Extended Data Fig. 1a-e and Supplementary Tables 1 and 2). Further gene mapping revealed that QTgw.cau-4B was associated with deletion of a fragment of about 500 kilobases, designated as r-e-z, in the Heng genome (Fig. 1a), as observed in a previous study 16 . The r-e-z fragment deletion resulted in the loss of three high-confidence genes, Rht-B1, EamA-B and ZnF-B (Extended Data Fig. 1f). Further genetic analysis confirmed that the genotypes with the r-e-z deletion showed a similar effect in shaping semi-dwarfism as the genotypes carrying the Rht-B1b, EamA-B and ZnF-B alleles ( Fig. 1b and Extended Data Fig. 2b). However, the deletion of the r-e-z haploblock was strongly associated with higher grain weight when compared to that of the genotypes carrying the Rht-B1b, EamA-B and ZnF-B haploblock, indicating a potential application of r-e-z haploblock deletion in enhancing the grain yield of semi-dwarfing varieties ( Fig. 1b and Extended Data Fig. 2a,b). The highly conservative genomic sequence of the r-e-z haploblock among wheat accessions and in other plant species (Extended Data Fig. 1g,h) indicates potentially broad applications of r-e-z block deletion in designing new semi-dwarf varieties of wheat and other crops.

r-e-z confers desirable semi-dwarf trait
To assess the phenotypic effects of the r-e-z haploblock deletion, we generated a pair of near-isogenic lines (NILs) with NIL-Heng harbouring the r-e-z deletion and NIL-Shi carrying Rht-B1b, EamA-B and ZnF-B in chromosome 4B. Both NILs carry Rht-D1a in 4D and showed similar plant height, but NIL-Heng showed more favourable agronomic traits, including more compact plant architecture, thicker and sturdier culms, larger flag leaves and spikes, and higher grain weight than NIL-Shi ( Fig. 1b and Extended Data Fig. 2a-f). NIL-Heng also showed significantly improved NUE as evidenced by its higher biomass under the low-nitrogen condition and higher NO 3 uptake rate than those of NIL-Shi (Fig. 1c,d and Extended Data Fig. 2g-i). The degree of the NUE improvement was positively correlated with GRF4 protein levels in NIL-Heng (Fig. 1e), most likely owing to reduced DELLA protein levels 2 . Overall, these improved traits conferred by the r-e-z deletion resemble an ideal plant architecture towards sustainable wheat production by shaping wheat plants with reduced tiller numbers, large spikes, thick and sturdy stems, and improved NUE as described in rice 3,17 . r-e-z enhances grain yield in semi-dwarf wheat Field tests of the two NILs in preliminary field trials planted at low and high densities (1.5-m-long rows) revealed that NIL-Heng produced higher harvest index, grain weight and grain yield, longer spikes and better culm quality than NIL-Shi at both low and high planting densities (Fig. 1f,g and Extended Data Fig. 2c-f). Notably, NIL-Heng exhibited a higher rate of increase in grain yield per unit than NIL-Shi as planting density increased (about 8.4% in low density, and about 11.9% in high density; Fig. 1g), suggesting superior adaptation of NIL-Heng to dense planting. In standard wheat field trials, the yield of NIL-Heng increased 12.1%, ranging from 10.6% to 13.8% at different planting densities, compared with those of NIL-Shi (Fig. 1h), which illustrates great potential of using r-e-z deletion to enhance grain yield of semi-dwarfing varieties.
Notably, severe plant lodging was found in the NIL-Shi plots, but not observed in the NIL-Heng even for the plots with high planting densities ( Fig. 1h and Extended Data Fig. 2j), suggesting that use of r-e-z deletion may also enhance yield stability.
Genotyping of a global collection of 556 wheat accessions identified the r-e-z deletion haploblock in only 12 Chinese wheat accessions   (Supplementary Table 3), indicating scarcity of the r-e-z deletion in modern wheat. Moreover, most of these r-e-z-deleted wheat accessions showed significantly higher TGW and larger spikes, but similar plant height, compared to those genotypes carrying Rht-B1b or Rht-D1b alleles (Extended Data Fig. 3).

Antagonistic effects between ZnF-B and Rht-B1b
To determine the gene(s) in the r-e-z haploblock responsible for the change in plant height and TGW, we created three independent mutants, znf-bb, eama-bb and rht1-bb, by gene editing of a semi-dwarf wheat variety, Fielder, to knock out ZnF-B, EamA-B and Rht-B1b alleles, respectively, on chromosome 4B (Extended Data Fig. 4a-c). Fielder has the same alleles at the three genes as in NIL-Shi, in which the Rht-B1b allele shows a strong suppressive effect on culm elongation and grain enlargement. The edited rht1-bb mutant was 14.22 cm taller and had a 5.59 g higher TGW (Fig. 2a,d and Extended Data Fig. 5a-c) whereas the znf-bb mutant was 8.40 cm shorter and had a 1.74 g lower TGW than Fielder (Fig. 2b,d and Extended Data Fig. 5d). Two EamA mutants (eama-bb and eama-aabbdd) showed similar plant height and TGW to those of Fielder (Extended Data Fig. 5e,f). These phenotypic data strongly support that in the r-e-z-deleted plants, the losses of ZnF-B and Rht-B1b conferred the semi-dwarf and increased TGW, respectively. In addition, Rht-B1b deletion (rht1-bb) resulted in a marked increase in plant height, spike length and TGW, whereas ZnF-B deletion (znf-bb) led to a slight reduction in grain size and plant height with no change in spike length compared to those of Fielder ( Fig. 2a,b). The znf-bb rht1-bb double mutant showed similar plant height to that of Fielder but longer spike and larger grain size than those of Fielder (Fig. 2c,d and Extended Data Fig. 4d). Therefore, the ZnF-B deletion confers a similar semi-dwarf trait to that of Rht-B1b, but less pleotropic effects on grain traits than Rht-B1b and has a great potential to replace the green revolution genes in semi-dwarf wheat breeding.

ZnF is a positive regulator for BR signalling
As ZnF regulates plant height, we further explored its biological functions by evaluating the phenotypic changes of the edited ZnF mutants. The znf-bb mutant produced shorter coleoptiles in the dark (Extended Data Fig. 6a), and showed much lower sensitivity to the application of epi-brassinolide (eBL, the active BR) than Fielder (Extended Data Fig. 6b), which is consistent with the observation of the BR-insensitive phenotype in NIL-Heng (Fig. 3a). To rule out functional redundancy from ZnF homoeologues, we generated znf-aabbdd triple mutants by knocking out all three ZnF homoeologues from the A, B and D subgenomes (Extended Data Fig. 4a). As expected, all of the znf-aabbdd mutants had significantly shorter coleoptiles in the dark and plant height, and were insensitive to eBL and brassinazole, a BR biosynthetic inhibitor ( Fig. 3b- Table 4). These results indicated that ZnF may act as a positive regulator for BR signalling. The epistatic interaction of BR signalling with GA biosynthesis as previously reported 11,18,19 was also observed in the znf-aabbdd mutants. The bioactive GA biosynthetic gene DWARF18(D18) encoding GA3-oxidase-2 was significantly downregulated, whereas the bioactive GA deactivation genes, GA2ox10 and GA2ox3, were upregulated (Extended Data Fig. 6g), resulting in a reduction in endogenous bioactive GA levels in the mutants (Extended Data   Fig. 6i,j). Meanwhile, the levels of endogenous BR, including castasterone and typhasterol, were not significantly different between the znf-aabbdd mutants and Fielder (Extended Data Fig. 6k).

ZnF-TaBRI1-TaBKI1 module gates BR signalling
ZnF is an evolutionarily conserved gene across plant species and is orthologous to Thermo-tolerance 3.1 (TT3.1) in rice 20 (Extended Data Fig. 8a). ZnF harbours a coiled coil domain and a RING-finger domain in its carboxy terminus (CT) and seven transmembrane domains in its amino terminus, suggesting that ZnF is a plasma membrane (PM)-localized protein (Extended Data Fig. 8b I   I   II   II   III   III   IV   IV  suppresses this perception 12,21 . To determine whether ZnF is functionally related to these PM-localized BR signalling regulators, we isolated wheat orthologues of BRI1, BAK1 and BKI1 (Extended Data Fig. 9a,b) and investigated their physical interactions with ZnF. The results confirmed that ZnF specifically interacted with TaBKI1 ( Fig. 3f-h) and TaBRI1 (Fig. 3j,k), but not with TaBAK1 (Extended Data Fig. 9d). Moreover, eBL enhanced ZnF-TaBKI1 interaction (Fig. 3i), but reduced ZnF-TaBRI1 conjugation (Fig. 3l). The addition of TaBRI1 intensified ZnF-TaBKI1 interaction ( Fig. 3m and Extended Data Fig. 9e). These results confirm that TaBRI1, TaBKI1 and ZnF together form a dynamic BR-responsive protein complex in which TaBRI1 facilitates the ZnF-TaBKI1 conjugation in response to BR signalling.

ZnF degrades TaBKI1 on the PM
Most RING proteins function as E3 ubiquitin ligases to trigger protein ubiquitylation and degradation 22 . The znf-aabbdd mutant expressed a higher level of TaBKI1, but the same level of TaBRI1, compared to those in Fielder ( Fig. 4a and Extended Data Fig. 9c), suggesting that ZnF might selectively degrade TaBKI1 in Fielder. In Nicotiana benthamiana cells, ZnF strongly suppressed TaBKI1 accumulation, but this was reversed after addition of a 26S proteasome inhibitor MG132 (Fig. 4b).
In a cell-free degradation assay, His-TaBKI1 was degraded faster in the protein extracts of Fielder than in the znf-aabbdd mutant ( Fig. 4c and Extended Data Fig. 9f). ZnF also ubiquitylated TaBKI1 both in vitro and in vivo (Fig. 4d,e and Extended Data Fig. 9g,h). Taken together, these results confirm that ZnF acts as an E3 ubiquitin ligase to ubiquitylate TaBKI1 for proteasomal degradation. Previous studies demonstrated that BR can trigger rapid dissociation of BKI1 from the PM into the cytosol, which defines a crucial mechanism underlying the fast elimination of PM-associated BKI1 to activate BRI1 (refs. 12,21,23,24). However, eBL quickly reduced the level of PM-associated TaBKI1-GFP fusion proteins only in Fielder protoplast cells, not in the znf-aabbdd mutant cells (Fig. 4f), indicating that the ZnF-mediated TaBKI1 degradation is required for the reduction of PM-associated TaBKI1 in response to the BR signal. We substituted amino acids in TaBKI1 to generate constitutively PM-associated TaBKI1(S208/212A) and TaBKI1(Y153F) and constitutively PM-disassociated TaBKI1(Y153D) protein mutants 23,24 ( Fig. 4g and Extended Data Fig. 9a), and found that ZnF selectively degraded   . e, The expression levels of BR metabolic and signalling genes in the znf-aabbdd mutant and Fielder measured by qRT-PCR (n = 3 biologically independent samples). Data in a-e are mean ± s.e.m. f-h, Interaction between ZnF and TaBKI1 confirmed by firefly luciferase (LUC) complementation imaging (f), bimolecular fluorescence complementation (g) and co-immunoprecipitation (co-IP; h) assays. i, eBL treatment enhanced ZnF-TaBKI1 interaction. j,k, Interaction between ZnF and TaBRI1 confirmed by firefly luciferase complementation imaging ( j) and co-IP (k) assays. l, eBL treatment (5 μM) attenuated ZnF-TaBRI1 interaction. m, Co-IP assay confirmed that TaBRI1 enhanced the interaction between ZnF with TaBKI1. EV, empty vector. Protein levels in i,l,m were quantified using ImageJ software (n = 3 independent experiments; data are mean ± s.d.). Arrowheads in b-d indicate the tips of coleoptiles (b,c) or main roots (d). Different letters in a,i indicate significant differences (P < 0.05, one-way ANOVA, Tukey's HSD test). In b-e,l,m, *P < 0.05; **P < 0.01; NS, not significant (two-tailed Student's t-test). In f-h,j,k, all experiments were repeated independently at least twice with similar results. Scale bars, 0.5 cm (a), 1 cm (b), 5 cm (c,d) and 50 μm (g).

Application of r-e-z deletion in wheat breeding
To introduce the r-e-z haploblock deletion into the GRVs that are grown at present in commercial production to obtain new semi-dwarf wheat varieties with enhanced grain yields, we crossed Nongda4803 (ND4803) harbouring Rht-B1b and wild-type Rht-D1a and Erwa carrying the r-e-z deletion and wild-type Rht-D1a and selected the r-e-z deletion block using markers and other traits using conventional phenotypic selection methods (Fig. 5a) in the breeding population. Finally, we successfully selected four lines (Q69, Q70, Q72 and Q84) with desirable plant height and yield. In a field trial, these lines showed yield increases of 6.48% to 15.25% compared to the control Liangxing99 (LX99), a Rht-D1b high-yielding variety widely grown in China with cumulative planting area exceeding 5 million hectares ( Fig. 5b and Table 1). The yield increase in these r-e-z-introgression lines was mainly attributed to marked increase in grain number per spike and TGW in comparison with the LX99 control, although the r-e-z-introgression lines had lower spike number per unit area than LX99 (Table 1), revealing different yield component profiles between the r-e-z-introgression lines and traditional GRVs. Taken together, these findings illustrate that our newly designed wheat breeding system that uses the r-e-z haploblock deletion to achieve semi-dwarfism not only effectively reduces plant height like Rht-B1b, but also increases yield potential and sustainability of wheat production.

Discussion
Since the 1960s, the NUE-repressing alleles (Rht-B1b and Rht-D1b) have been present in almost all commercially grown wheat varieties worldwide, which has created a substantial challenge to global sustainable wheat production due to increased N fertilizer input requirement 2,5,25,26 . In this study, we identified a natural r-e-z haploblock deletion that results in the loss of three genes, Rht-B1, EamA-B and ZnF-B. Compared to Rht-B1b lines, the lines with r-e-z haploblock deletion conferred the same semi-dwarf trait, but with considerably higher NUE, more compact plant architecture, larger spikes and grains, higher grain yields, and a more stable population suitable for dense planting (Fig. 1c-h and Extended Data Fig. 2). The higher accumulation of GRF4 protein and lower abundance of DELLA protein in NIL-Heng harbouring the r-e-z deletion than in NIL-Shi carrying the Rht-B1b allele (Fig. 1e) suggested an antagonistic interaction between GRF4 and DELLA 2 . Notably, this r-e-z haploblock deletion is very rare in modern wheat accessions, and thereby can be readily deployed into new wheat varieties to break the grain yield ceiling resulting from widespread application of the green revolution alleles, as demonstrated in this study (Fig. 5 and Table 1).
The data from gene editing of Fielder demonstrated that the divergent roles of the r-e-z deletion in reducing plant height and increasing grain weight and NUE are attributed to the combined effects of deletion of both Rht-B1 and ZnF-B (Fig. 2c), thus defining the two neighbouring genes as an integral genetic unit for fine-tuning multiple agronomic and yield traits in wheat (Extended Data Fig. 10). Unlike the gain-of-function Rht-B1b (or Rht-D1b) allele that strongly represses not only culm elongation but also spike and grain development and NUE 2,5,25 (Fig. 2a,c,d), ZnF  is probably a plant height-regulating gene whose null allele (illustrated by znf-bb) confers a semi-dwarfing effect with no or little undesired pleotropic effects on other agronomic traits (Fig. 2b-d and Extended Data Fig. 10a). Thus, we propose a new strategy to redesign semi-dwarf varieties by deleting the widely used Rht-B1b to overcome the growth defect and yield penalties caused by this green revolution allele and ZnF-B to retain the semi-dwarf statures. This can be achieved through genetic engineering such as the genotype-independent CRISPR-Cas9-based multigene-editing strategy in wheat 27,28 . Mechanistically, the semi-dwarf trait conferred by ZnF deletion is due to BR signalling deficiency, which is largely different from the traditional GA-insensitive semi-dwarfism induced by Rht-B1b or Rht-D1b. At a molecular level, ZnF acts as an E3 ligase to specifically target TaBKI1 for proteasomal degradation, and thus facilitates BR perception (Extended Data Fig. 10b,c); loss of ZnF dampens the BR-triggered TaBKI1 elimination from the PM, leading to a BR-deficient semi-dwarfism (Extended Data Fig. 10d). This ZnF-mediated regulation of BR signalling should be highly conserved across monocots and dicots, and further work will elucidate ZnF gene functions in other crops such as rice and maize. Thus, we may expand the application of ZnF as a new source of semi-dwarfing genes to breed new high-yielding varieties with desired plant height by reducing BR signalling in different crops. In summary, our study not only provides a new strategy to improve GRVs by engineering a functionally independent but genetically linked ZnF-DELLA genetic factor for sustainable agriculture, but also reveals a vital molecular mechanism of full degradation of the PM-localized BR receptor for effective activation of BR signalling.

Online content
Any methods, additional references, Nature Portfolio reporting summaries, source data, extended data, supplementary information, acknowledgements, peer review information; details of author contributions and competing interests; and statements of data and code availability are available at https://doi.org/10.1038/s41586-023-06023-6.  Publisher's note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.

Plant materials and growth conditions
An F 2 population of 286 plants was initially generated from a cross between a low-TGW parent, Shi4185 (Shi), and a high-TGW parent, Heng597 (Heng), and used to identify QTgw.cau-4B for TGW. To finely map QTgw.cau-4B, phenotypic and marker screening of the recombinants from F 3 to F 7 generations, coupled with phenotypic evaluation of the progenies, identified a key residual heterozygous line, R4, that showed heterozygosity within the interval of QTgw.cau-4B. NIL-Shi and NIL-Heng were evaluated for agronomic traits in the field at the China Agriculture University-Jize Experimental Station. The preliminary yield trial was conducted in the 2021-2022 growing season. Two NILs were hand planted in 30-row plots of 1.5 m in length with 90 seeds per row. The space between each row was 0.3 m for low-density and 0.15 m for high-density planting, with nine replicates. The two NILs were also planted in standard yield trials in 1.2 × 7 m plots using a planter in 2022. The experiment used a paired-plot design with three planting densities and ten replicates.
Breeding lines Q69, Q70, Q72 and Q84 containing the r-e-z haploblock deletion and Rht-D1a allele were selected in a field experiment at the National Observation and Research Station of Agriculture Green Development (Quzhou county, Handan city, Hebei province, People's Republic of China). Two elite breeding lines, Nongda4803 (ND4803, harbouring Rht-B1b and wild-type Rht-D1a as the female parent) and Erwa (with r-e-z deletion and wild-type Rht-D1a as the male parent) were used to develop the breeding population. During the 2018-2019 growing season, we phenotyped and genotyped more than 2,000 F 2 plants and obtained 91 outstanding plants carrying the combination of r-e-z block deletion and desirable agronomic traits. The selected individuals were further selfed to generate independent F 3 progeny, and phenotyping and genotyping of the F 3 lines identified five lines with the r-e-z block deletion and uniform appearance in the 2019-2020 season. The selected F 3 rows were bulk harvested to form F 4 for uniformity and yield performance evaluation in 2020-2021 field plots by planting them in a standard field trial with 1.2 × 7 m plots at a planting density of 3.3 million seedlings per hectare. A high-yielding GRV, LX99, was used as the yield control.

Field trait evaluation
Wheat seeds were randomly sampled from preliminary yield trials and standard field trials to measure TGW, grain length, grain width, grain aspect ratio and grain roundness using a Wanshen SC-G seed detector (Hangzhou Wanshen Detection Technology). The other agronomic traits including spike length, grain numbers per spike and flag leaf morphology were measured manually before harvest in the field. A digital dynamometer (YLK-500, ELECALL) was used to measure the bending strength of the fourth internode (from top to bottom). To assess the final yields of the NILs in either preliminary yield trials or standard field trials under different planting density, one 1-m 2 area (1 × 1 m) was randomly selected in each plot, and all wheat plants within the selected area in the plot were harvested. Before harvesting, the plants outside the selected 1-m 2 area were removed to avoid margin effects.
For field trait evaluation of r-e-z-carrying breeding lines, the plants within the standard field plots were all harvested for final yield evaluation. TGW was calculated from 3 randomly selected samples per plot with 500 grains in each sample. Grain number per spike was counted manually from 3 randomly selected replicates of 20 main spikes in each plot. Spike number per unit area was assessed by counting all of the spikes within a randomly selected 1-m-long row, and 3 replicates in each plot were carried out.

QTL mapping and gene cloning
Single sequence repeat markers were screened in the F 2 population of Heng × Shi to map the QTLs for grain traits. One major QTL (QTgw.cau-4B) for TGW, grain length and grain width was located on the short arm of chromosome 4B and three single sequence repeat markers were mapped within the interval of QTgw.cau-4B. Further genotypic and phenotypic analyses of the F 3 -derived residual heterozygous line mapped QTgw.cau-4B to the interval between the markers SNP-5 and SNP-7. The QTL explained 74.65% of the phenotypic variance. Recombinants between the flanking markers of QTgw.cau-4B were continuously screened from F 4 to F 7 generations for fine mapping, and phenotypic and genotypic data from the recombinants narrowed the QTL interval to the region between the markers M7 and ID-51 where only six high-confidence genes were annotated on the basis of RefSeq v1.1 (2018) produced by the International Wheat Genome Sequencing Consortium. Both Shi and Heng were resequenced, and their genomic sequences in the QTL region were compared to identify sequence polymorphisms. The primers used for map-based cloning are listed in Supplementary Table 5.

Plasmid construction
For the firefly LUC complementation imaging (LCI) assay, the full-length coding sequences (CDSs) of the candidate genes, including ZnF, TaBKI1, TaBRI1 and TaBAK1, were separately cloned into the pCAMBIA1300-nLUC and pCAMBIA1300-cLUC vectors through In-Fusion PCR cloning system (CL116, Biomed). For the bimolecular fluorescence complementation (BiFC) assay, CDSs for TaBKI1 and ZnF were cloned into the pEarleygate201 and pEarleygate202 vectors using a Gateway cloning system (12535029, Invitrogen). For the co-IP assay, the ZnF-MYC, BKI1-GFP, BRI1-GFP and BRI1-Flag constructs were generated by inserting the CDSs of these genes into pCAMBIA1300 vectors fused with different tag sequences (MYC, GFP and Flag) using an In-Fusion PCR Cloning kit (CL116, Biomed). To generate His-BKI1 and MBP-ZnF-CT (314-473 amino acid) constructs, we used pCold-TF (fusing with His tag; Takara) and pMAL-c2X (fusing with maltose binding protein tagged) vectors. All of the primer sequences are listed in Supplementary Table 6.

CRISPR-Cas9-mediated gene editing
The CRISPR-Cas9-based gene editing was used to knock out target wheat genes. The single guide RNA (sgRNA) target sequences were designed according to the exon sequences of the target genes using the online software E-CRISPR (http://www.e-crisp.org/E-CRISP/). The MT1T2 vector was amplified using the primers containing sgRNAs and then cloned into the CRISPR-Cas9 vector pBUE411. The generated vector was further transformed into the Fielder variety following the Agrobacterium tumefaciens (strain EHA105) gene transformation procedure 29 . Subgenome-specific primer pairs were designed for mutation analysis and further screening of homozygous T 2 and T 3 mutant lines (Supplementary Table 6). eBL and brassinazole treatment eBL (E1641, Sigma) and brassinazole (BRZ; B2829, TCI) were separately prepared by dissolving them in dimethylsulfoxide (DMSO). For eBL or BRZ treatment, 2-day-old wheat seedlings of Fielder and a znf-aabbdd mutant (line 2) were soaked in defined concentrations of eBL or BRZ water solution. The same volume of DMSO (blank solvent) was used as a mock control. The lengths of coleoptiles and roots were determined 7 days after the treatments. For the lamina joint inclination assay in response to eBL treatment, 1.5-cm-long leaf segments containing lamina joints were excised from 14-day-old seedlings of NIL-Shi and NIL-Heng, and incubated in eBL solutions at different concentrations in the dark for 2 days. The lamina joint angles were determined by ImageJ software (https://imagej.net/ij/). All experiments were repeated three times.

Histological analysis
The middle part of the first internode (from top to bottom) at the heading stage (emergence of inflorescence completed at Zadoks stage 58) and developing grain at 10 days after pollination 30 were collected to determine cell size. The collected samples were fixed in an FAA solution (10% (v/v) formaldehyde, 50% (v/v) alcohol, 5% (v/v) acetic acid and 35% (v/v) water) overnight at 4 °C, and then were embedded in paraffin, dehydrated and decolourized as described previously 31 . The samples were then cut into 4-μm-thick cross-sections using a Leica Ultracut rotary microtome (Leica Biosystems), and stained with periodic acid Schiff or 1% sarranine and 0.5% fast green (G1031, Servicebio). Photographs were taken with a microscope imaging system (DS-U3, Nikon) and the cell lengths were measured with CaseViewer 2.3 (3DHISTECH).

qRT-PCR and RNA-sequencing assays
For the qRT-PCR assay, total RNA was extracted from wheat tissues using a TRIzol reagent (Thermo Fisher Scientific) following the manufacturer's instructions. After the removal of genomic DNA, cDNAs were synthesized using a Reverse Transcription kit (R223, Vazyme). Real-time PCR was carried out using the SYBR Green PCR Master Mix (Q121, Vazyme) in a CFX96 Real-Time PCR System (Bio-Rad). β-ACTIN was used as the internal gene control. Each experiment was repeated three times. The primers used for qRT-PCR assays are listed in Supplementary Table 6.
For RNA-sequencing analysis, the stem samples were collected at the jointing stage (second node detectable at Zadoks stage 32) 30 , and total RNAs were extracted using TRIzol reagent. The cDNA libraries were constructed using Poly-A Purification TruSeq library reagents (Illumina), followed by sequencing on an Illumina 2500 platform. After cleaning up raw sequence reads, the clean reads were mapped to the wheat reference genome (International Wheat Genome Sequencing Consortium, RefSeq v1.1) using TopHat2 software 32 . The differentially expressed genes were analysed using the DESeq2 R package. Significant differentially expressed genes were determined using a standard procedure including adjusted P value (false discovery rate < 0.05) and fold change ratio (log 2 [FC] ≥ 1). Gene ontology enrichment was carried out using the online tool Triticeae-GeneTribe 33 .

Antibody preparation
To create the anti-TaBKI1 antibody, a peptide fragment, N-EGRDDTAGKAEEDRK-C, corresponding to amino acids 121-135 of TaBKI1 was artificially synthesized and purified, and then conjugated with the keyhole limpet haemocyanin carrier before generation of the anti-TaBKI1 antibody in a rabbit.

LCI and BiFC assays
The LCI and BiFC assays were carried out in N. benthamiana leaves. In brief, the nLUC and cLUC derivatives, or the nGFP and cGFP derivatives, were transformed into the A. tumefaciens strain GV3101. The obtained Agrobacteria cells harbouring the constructs were co-infiltrated into N. benthamiana leaves. The LUC activity was analysed 48 h after infiltration using Night SHADE LB985 (Berthold), and the fluorescence signal of GFP was observed 48 h after infiltration under a confocal microscope (LSM880, Zeiss).

Protein subcellular localization
To localize ZnF proteins in a cell, the 35S::ZnF-GFP and 35S::PIP2-RFP expression vectors were separately transformed into the A. tumefaciens strain GV3101, and then were co-expressed in the leaf epidermal cells of N. benthamiana. The GFP fluorescence signal was detected about 48 h after infiltration by a confocal microscope (LSM880, Zeiss). For the assays using protoplast cells, wheat protoplasts were initially isolated from the first leaf of 7-day-old seedlings, and then the 35S::ZnF-GFP or 35S::TaBKI1-GFP expression vectors were separately transferred into protoplast cells following the protocol described previously 35 . The GFP fluorescence signal was detected 16 h after the transformation by a confocal microscope (LSM880, Zeiss).

Cell-free degradation assays
Total proteins were extracted from 2-week-old wheat seedlings with native buffer (50 mM Tris-MES pH 8.0, 0.5 M sucrose, 1 mM MgCl 2 , 10 mM EDTA, 5 mM dithiothreitol, 1 mM PMSF and 1× protease inhibitor cocktail) 37 . The protein extracts of Fielder or the znf-aabbdd mutant were mixed with purified His-BKI1 fusion protein in the presence or absence of 50 μM MG132. The samples incubated at room temperature (25 °C) were collected at designated time points, followed by the addition of 2× SDS loading buffer to stop the reaction. The proteins were detected by SDS-PAGE using anti-His (1:2,000 dilution, BE2017, EASYBIO) antibody.

Phylogenetic, genetic diversity and micro-collinearity analyses
The protein sequences of ZnF and its orthologues from different plant species were extracted from the EnsemblPlants database (http://plants. ensembl.org/index.html). The phylogenetic tree was constructed using a maximum-likelihood method in the MEGA5.0 program with bootstrap (500 replicates) and complete deletion. Wheat accessions including 28 wild emmer accessions, 93 domesticated tetraploid wheat accessions and 289 hexaploid wheat accessions (Supplementary Table 7) were used for the nucleotide diversity analysis of the Rht-B1, EamA-B, ZnF-B gene cassette and its flanking regions using VCFtools (v0.1.13) with >100-kilobase sliding windows in 100-kilobase steps. The online tool Triticeae-GeneTribe was used for micro-collinearity analysis of the Rht-B1, EamA-B, ZnF-B gene cassette among different crop species 33 .

Hydroponic cultivation for low-nitrogen treatment
The wheat seeds were initially germinated on wet filter papers. About 3 days later, the seedlings were transferred to hydroponic culture (2.5 mM KNO 3

15
N uptake rate analysis For 15 N uptake analysis, wheat seedlings were cultured in the hydroponic culture (supplemented with 2.5 mM KNO 3 , pH 5.8) for two weeks. After N starvation by culturing the seedlings in a hydroponic solution without N for 2 days, wheat roots were treated with K 15 NO 3 (98 atom% 15 N; SigmaAldrich, number 335134) for 30 min. After washing with 0.1 mM CaSO 4 solution and deionized water as described previously 38 , roots of seedlings were collected and dried at 70 °C for 3 days. After grinding the sample into powder, the 15 N content in the root was measured using an isotope ratio mass spectrometer (Thermo Finnigan Delta Plus XP; Flash EA 1112) with three biological replicates.

Endogenous phytohormone quantification
The stem tissues of the indicated wheat materials were collected at the early jointing stage (second node detectable at Zadoks stage 32) 30 , and were immediately frozen in liquid nitrogen. The quantification of endogenous GAs and BRs levels was carried out as reported previously 39 . In brief, 200 mg of the ground plant material powder was extracted with 90% aqueous methanol. Simultaneously, each of the D-labelled GA and BR compounds was added to the extraction solvents as internal standards for quantification. After effective pretreatment, GA and BR analysis was carried out on a quadrupole linear ion trap hybrid mass spectrometer (QTRAP 6500, AB SCIEX) equipped with an electrospray ionization source coupled with an ultrahigh-performance liquid chromatography instrument (Waters).

Reporting summary
Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Data availability
The raw sequence data generated by this research have been deposited in the National Center for Biotechnology Information under the accession number PRJNA852953 for RNA sequencing. The raw sequence data of previously published resequenced accessions used in this study are available in the Sequence Read Archive under the accession codes PRJNA597250, PRJNA439156, PRJNA663409, PRJNA596843 and PRJNA544491. All other data are available in the main text or the Supplementary Information. Source data are provided with this paper. indicate the insertion and deletion of nucleotide, respectively, and the base numbers of insertion/deletion (bp) were shown on the right. The sgRNA target sequences were marked by pink boxes, and the PAM motifs were highlighted in red letters.

Extended Data Fig. 5 | Simultaneous mutations in ZnF-B and Rht-B1b in
Fielder mimic the effect of r-e-z deletion. a, Comparison of the agronomic traits including spikelet number per spike, fertile spikelet number, grain number per spike, grain length and grain width between Fielder (harboring Rht-B1b) and rht1-bb single mutant. b, Comparison of cell lengths of longitudinal section of the fully elongated uppermost internodes collected at anthesis stage between Fielder and the rht1-bb mutant (n = numbers of parenchymatic cells). Scale bar, 20 μm. c, Comparison of cell lengths of the cross sections of the developing grains collected at 10 days after pollination between Fielder and rht1-bb mutant (n = numbers of pericarp cells). Scale bars are 500 μm for the upper panels, and 100 μm for the lower panels. d, Comparison of the developmental and yielding traits between Fielder and znf-bb single mutant. e,f, Comparison of plant height, spike morphology, and grain traits between Fielder and two Eama mutants: eama-bb (e) and eama-aabbdd (f). In a,d,e,f, n = numbers of biologically independent samples. In a-f, data are means ± s.d.; P values were calculated by a two-tailed Student's t-test (** P < 0.01; * P < 0.05; ns, no significant difference); Scale bars are 10 cm for whole plant, 5 cm for spike, and 1 cm for grain.

March 2021
Corresponding author(s): Zhongfu Ni Last updated by author(s): Mar 5, 2023 Reporting Summary Nature Portfolio wishes to improve the reproducibility of the work that we publish. This form provides structure for consistency and transparency in reporting. For further information on Nature Portfolio policies, see our Editorial Policies and the Editorial Policy Checklist.

Statistics
For all statistical analyses, confirm that the following items are present in the figure legend, table legend, main text, or Methods section.

n/a Confirmed
The exact sample size (n) for each experimental group/condition, given as a discrete number and unit of measurement A statement on whether measurements were taken from distinct samples or whether the same sample was measured repeatedly The statistical test(s) used AND whether they are one-or two-sided Only common tests should be described solely by name; describe more complex techniques in the Methods section.
A description of all covariates tested A description of any assumptions or corrections, such as tests of normality and adjustment for multiple comparisons A full description of the statistical parameters including central tendency (e.g. means) or other basic estimates (e.g. regression coefficient) AND variation (e.g. standard deviation) or associated estimates of uncertainty (e.g. confidence intervals) For null hypothesis testing, the test statistic (e.g. F, t, r) with confidence intervals, effect sizes, degrees of freedom and P value noted

Software and code
Policy information about availability of computer code Data collection The fluorescence signal was detected using a confocal microscopy (LSM880, Zeiss).
Images from immuno blotting were collected with CLINX (ChemiScope 6000). The LUC activity was analyzed using the Night SHADE LB985 (Berthold). Illumina NovaSeq platform was used to collect the sequencing data. Bio-Rad CFX96 with CFX Maestro 1.1 software was used to qPCR analysis. The 15N content was measured using an isotope ratio mass spectrometer (Thermo Finnigan Delta Plus XP; Flash EA 1112). For histological analysis, photographs were taken with a microscope imaging system (DS-U3, Nikon, Japan) and the cell lengths were measured with CaseViewer 2.3 (3DHISTECH, Ltd., Hungary). For endogenous phytohormone quantification, GAs and BRs analysis was performed on a quadrupole linear ion trap hybrid mass spectrometer (QTRAP 6500, AB SCIEX). For manuscripts utilizing custom algorithms or software that are central to the research but not yet described in published literature, software must be made available to editors and reviewers. We strongly encourage code deposition in a community repository (e.g. GitHub). See the Nature Portfolio guidelines for submitting code & software for further information.